Journal of Economic Growth, 4: 55–80 (March 1999)c© 1999 Kluwer Academic Publishers, Boston.

Mercantilism and Modern Growth

JOHN MCDERMOTT

Department of Economics, University of South Carolina, Columbia, SC 29208 USA

Nations close themselves voluntarily to varying degrees. Restrictions on the flow of ideas are difficult to understand,since open countries have higher relative incomes. This article provides an explanation based on the existence oftwo channels of public finance—traditional and mercantilistic. The latter refers to monopoly creation to providea stream of government revenue. Strong, profitable monopolies require that the nation be closed to new ideasabout technology and organization. The government sets the degree of restriction to balance current mercantilisticrevenue with future revenue from traditional sources. The model is supported with numerical simulations andhistorical illustrations.

Keywords:mercantilism, growth, taxation, openness, familiarity

JEL classification: F43, N10, O10, O38

To be precise [Colbert’s mercantile policy] was literally a kind of indirect taxation,taxing the consumers through the monopolistic artisans.. . . Industrial control thusstood revealed as naked fiscalism. (Eli Heckscher, 1935, pp. 180–182)

Mercantilism had six aspects, not five: the sixth was public (or royal) finance, andone might with great cogency maintain that it was the most important of the lot.(C. Herbert Heaton, 1937)

In the interpretation developed here, the supply and demand for monopoly rightsthrough the machinery of the state is the essence of mercantilism.. . . the statefound it efficient to seek revenues by selling monopoly and cartel (guild) privileges.(Mercantilism, conceived as a broad process of economic regulation, seems to bea perennial state of most societies (Robert Ekelund and Robert Tollison, 1981,pp. 5–6, 153)

1. Introduction

Postwar economic data allows us to make two general observations. First, thedistributionofworld per capita income relative to the world leader has shown little convergence, althoughit has not been completely stable. Second, openness appears to have a positive effect inraising anindividual nation’sper capita income. Both propositions have found support ina variety of empirical studies.1 In addition, note that openness is anabsolute, not a relative,measure. That is, it is possible for a country to be perfectly open or perfectly closed. Fromthis property it follows that a general shift to the right in openness would also entail a

56 MCDERMOTT

degeneration of its global distribution, as nations became crowded at the high end. Animportant implication of all this is the following: if the world’s countries as a group hadbecome increasingly open in the last decades, we would have seen more than very meagerconvergence in the distribution of per capita relative income. Openness appears to havebeen limited and resisted by many countries. Why is a policy that can confer important andlong-lasting benefits rejected so often?

The answer that I develop in this article begins with the observation that throughout historygovernments have had two sources of revenue, atraditional source and amercantilisticsource. By the former I mean the direct taxation of households; by the latter, I mean anindirect process of taxing householdsthroughmonopolies. By establishing a monopolythe monarch (or modern government) can create a single large stream of revenue that ismuch easier to tax than the several small streams of the individual households. To theextent that the government does rely on the mercantilistic tax source, it is in its interest toclose the economy—though rarely completely—to new ideas, technologies, and businessorganizations in order to protect the monopolies’ revenue flow. An inevitable consequenceof granting a monopoly right is the denial of rights to competitors, those in the best positionto advance the state of technical knowledge and general skill.

Mercantilism, as a theoretical system, has been impossible to define.2 It becomes moreintelligible as a system of fiscal expediency by which the government and certain favoredorganizations jointly extract income from the population. This view has been carefullyarticulated and illustrated by Ekelund and Tollison (1981), but traces of the argument go backquite far.3 It is not hard to find evidence (see Section 5) that a principal (but not exclusive)way to generate revenue was by creating and nurturing monopolies that would then cedea portion of their revenue to the government. Actual transfers to the state were made inmany ways, often through “loans” that were never repaid, at times as bribes, sometimesas dividends, sometimes as taxes. Once granted privileges, monopolies had a substantialinterest in limiting the spread of new knowledge about production and organization to theircompetitors. The guild system, for example, was famous for slowing the development ofnew techniques and stopping their use once they became known.

Stagnation, however, is not inevitable. Some countries have opened their economiesand reaped the benefits of growth in knowledge. Given the strength of the underlyingmercantilistic forces, I identify two secular processes that determine the degree to whichthe government opens its economy. First, growth in human capital—the very essence ofdevelopment—encourages the government toreduceopenness and increase its relianceon monopolies, since greater knowledge raises labor productivity. Second, however, de-velopment raises the traditional average tax rate, which leads the government to increaseopenness and partly reject monopoly collaboration. The balance of these forces determinesthe country’s place in the global distribution of income.

Cycles in openness are possible in the model because I assume that government canchoose policy only at discrete times. This constraint raises the possibility that the state willfind itself in an unpalatable position in which its best option is to become extremely closed.As a result, income falls considerably. But a consequence of the decline is that at its nextopportunity, it selects a far more open policy. Such cycles are typical in the developingworld (Krueger, 1993).

MERCANTILISM AND MODERN GROWTH 57

The model of this article relies on recent theoretical work highlighting the role of knowl-edge flows in raising growth or relative income. Rivera-Batiz and Romer (1991) comparethe growth rates of two economies when they were either completely separate of fully inte-grated. In the latter case, diffusion could be said to be instantaneous, and growth rises due toscale economies in ideas in generating knowledge. In Parente and Prescott (1994) the focusis on the barriers placed in the way of technological information from abroad. By slowingdiffusion, hindering the ability to accumulate technical and general knowledge, growthis reduced.4 The basic model here emphasizes technological receptivity in the manner ofGoodfriend and McDermott (1999). In contrast to previous work, however, the parameterthat governs the flow of knowledge here is endogenously determined by the government.

Douglass North (1981, chap. 2) saw the tradeoff between public revenue and efficientproperty rights as central to the theory of the state. Although he did not present a formalmodel of this tradeoff, he saw that a “ruler therefore frequently found it in his interest to granta monopoly rather than property rights which would lead to more competitive conditions”(p. 28). The theoretical structure presented below can be looked on as one way to formalizeNorth’s ideas. Robinson (1997) is also concerned with the reasons that governments mightvoluntarily restrict growth. In his theory, an autocrat might refuse to provide infrastructureto make it more difficult for the political opposition to mobilize against him. Rosenthal(1998) and North and Weingast (1989) explain the fiscal problems of seventeenth-centuryFrance and England in the context of the key constitutional conflict between the crown andthe aristocracy. The latter emphasizes a point that I make here: that monopoly is a terribleway to raise revenue; it is extremely detrimental for growth. Finally, Parente and Prescott(1998) construct a general-equilibrium model to explain why monopoly rights persist oncegranted and use numerical methods to estimate the size of the negative effect. The size oftheir effect—monopoly can reduce per capita income to a third of its free-market level—isconsistent with the estimates I get below. They do not, however, introduce a fiscal motivefor granting the rights in the first place nor do they consider growth explicitly.

The next two sections set up the basic model and derive the households’ equilibrium pathsfor work and knowledge accumulation. Section 4 describes the nature of the government’stax collection process. There I define precisely what is meant by traditional and mercan-tilistic revenue generation. Section 5 is historical, presenting evidence of mercantile financein the past and suggesting why it is still present today. Section 6 derives the main resultsusing numerical simulations. A final section offers a brief summary of those results.

2. Production and Accumulation

In this section, I build a simple model with the following properties: (1) in long-runequilibrium all countries grow at the same rate; (2) in long-run equilibrium the distributionof world income depends on the distribution of country openness to world technologicalideas.

The world consists of many small countries. In countryi , per capita output is given by

yi = Ähi eiW, (1)

58 MCDERMOTT

whereeW is the fraction of effort that the representative individual of the country devotesto working, h is the country-specific general knowledge that each possesses, andÄ is afactor that applies to all countries in the world. The variable isÄ as a measure ofworldtechnology(Parente and Prescott, 1994), which either has been accumulated as a result ofR&D investment made throughout history (Romer, 1990; Charles Jones, 1995), has arisenas a spillover from global specialization in production (Goodfriend and McDermott, 1995),was generated by learning by doing (Young, 1993; Lucas, 1993), or was created frompure increases in scale via population growth (Michael Kremer, 1993). World technologygrows exogenously. Although it is difficult to be precise about a measure of disembodiedproductivity, I think ofÄ as a stock of technical ideas that are available costlessly toproducers all over the world.

In addition to having access to world technology, individuals within each country possessa stock ofgeneral knowledgeor human capitalh that each person accumulates by theapplication of effort to learning. It has long been recognized that technical ideas are not putto use everywhere to the same extent. One reason for this may be that the general knowledgelevel h is so different across countries (Nelson and Phelps, 1966). Many techniques ofproduction could be used in developing nations if the general level of knowledge werehigher.5 This effect is captured in (1). General knowledge or human capital within acountry makes the world technology immediately more useful in producing goods.

Within any country, human capital arises according to the following:

h = h1−γ (κÄ)γeL . (2)

In this expression,eL is the fraction of effort spent learning. The country-specificκ liesbetween 0 and 1(0 < κ ≤ 1). This formulation captures the idea that there is aspilloverfrom the state of world technologyÄ to the creation of domestic human capital: peopleacquire general knowledge more readily the more sophisticated is the production processthey use. The variableκ, which we callfamiliarity, was first introduced by Goodfriendand McDermott (1998) and measures the degree to which the country has become opento, or familiar with, the outside world in the sense ofideas, not physical commodities.Familiarity measures the effectiveness of the spillover from the technology embodied inworld production in heightening individual learning productivity. A country is “perfectlyfamiliar” with the rest of the world ifκ = 1. If, on the other hand,κ = 0, then the economyis absolutely closed and learning does not take place. The parameterγ (where 0< γ < 1)measures the intensity of the technological spillover relative to the own-knowledge effect.Finally, we assume that human capital does not depreciate and that it is not costly to educatenew family members to provide them with the current level of knowledge.6

According to this technology, human capital grows at the rate

h

h≡ gh − κ

γ

FeL , (3)

where

F ≡(

h

Ä

)γ. (4)

MERCANTILISM AND MODERN GROWTH 59

The variableF—which is the principal state variable in the analysis to follow—measureshuman capital per unit of world technology.

3. Household Equilibrium

Households do not keep all of their earned income; as explained in the next section, afraction of income is transferred to the government and its agents. I assume that tax ratesare flat, so consumption is

c = ωheW, (5)

whereω < Ä. The expression forω, derived in the next section, is considered exogenousby the households.

The household decision-maker maximizes the discounted value of the utility of eachfamily member:

J =∫ ∞

0ln c(t)e−ρt dt, (6)

whereρ is the household rate of time preference. Instantaneous utility is logarithmic andfamily size does not influence utility. The maximization is subject to the motion equation(3) and the effort constraint:

1= eW + eL . (7)

For the allocation of work effort to be optimal it is necessary (see the appendix) that

eW = F

xκγ, (8)

wherex ≡ qhh and F is given in (4). The new variablex is the utility value of humancapital (qh is the shadow price). From the constraint (7), learning is given by

eL = 1− F

xκγ. (9)

Optimality also requires thatx move through time to satisfy

x

x= ρ + γ κ

γ

F− 1+ γ

x. (10)

Using (3), (4), and (9), we see thatF grows according to

F

F= γ

(κγ

F− 1

x− gÄ

), (11)

wheregÄ is the exogenous rate of growth ofÄ. Finally, all candidate paths ofF andxmust converge to constants.

60 MCDERMOTT

Figure 1. Household path to the steady state.

A solution to this problem has the form of an initial value forx, set by the representativehousehold. Since the initial value ofF is given, the motion equations above trace out theentire future progress of both variables. There is a unique value forx0 that causes the(x, F)pair to converge to the steady state.

To find the steady-state values, set the motion equations (10) and (11) to zero and solve

x∗ = 1

ρ + γgÄ, (12)

F∗ = κγ

ρ + (1+ γ )gÄ . (13)

All countries approach the same value ofx (it is assumed thatρ and γ are the sameworldwide) but the long-runF depends on the nation’s familiarity,κ. The steady-state pairis shown as point A in Figure 1. The effort allocations in equilibrium can then be foundfrom (8) and (9). SinceF is constant in the steady state, it must be true by (4) thath growsat the same rate asÄ : gh = gÄ. Per capita producty, as given by (1), will grow at thesum of the rates of growth ofh and the global productivity factorÄ, since work efforteW

is constant:

gy = gh + gÄ = 2gÄ. (14)

All countries grow at the same rate in the steady statedetermined by the pace of exogenoustechnical change in the world. It is not true, however, that the steady-statey is the sameeverywhere. The equilibrium value of a nation’s general knowledgeh is proportionallygreater the larger isκ. This follows directly from the definition ofF in (4) and the fact thatF∗ is proportional toκγ . Let yL stand for the per capita product of the lead country (withthe maximum familiarity of 1), use (4) to eliminatehi in (1), and note that in equilibriumeW

MERCANTILISM AND MODERN GROWTH 61

Table 1.Steady state.

Parameter Equilibrium Value

ρ 0.04gÄ = gh 0.01γ 0.20

gy 0.02eL 0.19

F∗/κγ 19.23κγ /F∗ 0.052

attains a common value across countries. From (13) it then follows that a nation’s relativeincome is given by

y

yL= κ. (15)

Familiarityκ actually shows how far away a nation is from the world leader in terms of percapita income. A country can catch up to world leaders only by raisingκ.

Table 1 provides a numerical example of the steady state. The first column lists parameters;the second shows endogenous variables. Growth in real output and consumption per capitais 2.0 percent, people devote 19 percent of their effort to learning, and the productivityof learning time,κγ /F∗, is 5.2 percent. None of these are affected by familiarity in thelong-run equilibrium.

In general, the initial value ofF will not be at its steady-state level. Sinceκ is fixed,Ä isgrowing exogenously, andh changes only slowly through the accumulation of knowledge,there will usually be a path of transitional growth inF before the long-run equilibrium isachieved. The differential equations (10) and (11) define a saddle-point equilibrium so thatif F begins below its long-run level, it will slowly rise to it: countries grow faster than theworld average if they begin with initial stocks ofF that are belowF∗. The saddle-pathis shown as QA in Figure 1. Each country converges to the value ofF determined by itsown particularκ, which means that each approaches a different per capita product path asillustrated below in Figure 2.

4. Mercantilism, Monopoly, and Government Finance

We now turn to the central question: why don’t governments simply open up their countriesby settingκ = 1? Such a strategy, as is clear from (15), would maximize per capita incomein the long run and allow the nation to grow fast enough in the short run to catch up to worldleaders.

We cannot ignore the fact that, in some cases, familiarity is extremely costly to raise.Geographic isolation may leave a legacy that makes integration into the world economydifficult even after many centuries. The linguistic development of China, for example,continues to inhibit the flow of information to and from the West. Low familiarity that arisesfrom such deep-seated causes is difficult to change. Yet I would argue that in most cases

62 MCDERMOTT

Figure 2. Income paths and familiarity.

familiarity could be raised at little cost simply by removing various “barriers to technologicalabsorption” (Parente and Prescott, 1994) and allowing more freedom to interact with the restof the world. If so, then it must be true that low familiarity is sometimes a rational strategy.One possibility is that closed economies occur where the traditional source of governmentfinance is weak. This forces the government, first, to create monopolies to generate a streamof revenue and, second, to protect these organizations by reducing familiarity. They erectbarriers to the flow of information in order to ensure that the monopolies are financiallyhealthy.

To bring the argument into focus, I assume that the government is concernedonlyabout itsrevenue and uses it essentially for consumption purposes. It is neither transferred back to thehouseholds nor used in public good provision that would increase output and tax revenues.Either feature could be built into the model, but since it would obscure the main point, Ichoose to ignore both possibilities.7 Given that revenue generation is the key element, Inow proceed to develop the three strands of the argument.

First, an outstanding feature of poor countries, both in the present and in the past, is therelatively small size of the middle class. Without a large middle class, traditional taxationis difficult. On the one hand, the rich are adept at hiding income and moving it out of thejurisdiction of the national government, making the effective cost of collection very high.On the other hand, poverty itself limits the average rate of taxation. Very little tax revenuecan be generated by traditional methods from that portion of the population that is poor.Below a certain income level, any taxation is inconsistent with survival, but even abovethat, the cost of monitoring and collection may be high enough to prohibit any meaningfulrevenue generation. To these reasons, we may add a historical one. In certain eras, revenuecollection by national governments was made extremely difficult by the claims of localjurisdictions, both towns and feudal political structures. For these reasons, I assume thatthe tax rate is largely exogenous. It rises with the level of development, but it cannot bechanged by the government.8

MERCANTILISM AND MODERN GROWTH 63

Second, through the ages governments have dealt with obstacles to revenue generation indifferent ways, including the debasing of the coinage, the use of tax farmers, and moderninflationary finance. I am going to focus on one particular method: chartering a monopolyto extract revenue from its profits. We know that governments explicitly granted monopolyrights during the mercantile era, and we observe many measures today that have a similareffect, although they are seldom so obvious as before. I hypothesize that these rights weregranted solely to create a stream of revenue that could be easily taxed. This does notnecessarily mean that the ill consequences of monopoly creation were ignored, only thatsuch effects were outweighed, in the mind of the sovereign, by the increase in revenue.

Third, if the mercantilistic source of revenue is important to the government, then weshould expect to see low familiarity, since monopolies are most healthy and profitable whenthe country is closed to competition in both goods and ideas from the outside world. Inother words, it is necessary for the government to place barriers in the path of competitors,both in terms of imports of commodities and of ideas, so that they do not challenge theposition of the monopolist. These restrictions translate into a lowκ, making learning costlyand keeping relative income low in the steady state (see (15)).

These ideas can be formalized as follows. First, assume that the monopolies succeed inextracting a fractions = σ(1− κ) of household incomey, whereσ is a positive constant.With this formulation, the more open a country, the smaller the transfer from householdto monopolist. The maximum transferσ occurs when the country is completely closed(κ = 0).9 From what remains of household income, the government taxes at thetraditionaltax rateτ , which is assumed to be constant. After both kinds of revenue extraction, percapita consumption is

c = (1− τ)(1− s)y = ωheW, (16)

whereω ≡ (1− τ)(1− s)Ä was used in the previous section in the household decisionproblem. Taxes are not allowed to reduce the level of income but merely to redirect it. Thisis a simplification that I justify on two grounds: because it keeps the analysis focused onthe question of openness, and because I have introduced an offsetting simplification in theform of not allowing the revenue collected by the government to be spent on public goodsto raise income.

Monopolies earn gross revenues of

RM = s y N= σ(1− κ)y N, (17)

whereN is total population in the country. The government taxes the monopolists’ grossreceipts at the ratem+τ , wherem> 0 is called themercantilistic tax premium. That is, thegovernment can tax the monopolies at a higher rate than it can tax the households.10 Thissource of revenue,(m+ τ)RM , added to the revenue obtained directly from households—which isτ(1− s)y N—yields total receipts to the government of

R= [τ +mσ(1− κ)]yN. (18)

Monopolies are valuable to the government because the income that they divert fromhouseholds is taxed at a greater rate than it could be if it remained with the households.

64 MCDERMOTT

5. Mercantilism: Past and Present

There can be little doubt that in the centuries before the industrial revolution a significantportion of government revenue in Europe came from the granting of monopoly rights.According to Ekelund and Tollison (1981, p. 6),

The state found it efficient to seek revenues by selling monopoly and cartel (guild)privileges. Such revenues supplemented the tax revenues available to the Englishand French monarchs, and indeed the degree of dependence on revenues frommonopolization was very significant in mercantile England and France. Duringthe administration of Colbert (1662–1683), for example, the French state procuredroughly one-half of its yearly revenues from the granting of monopoly and cartelrights.

If this quote is believable, mercantilistic finance was clearly important in the seventeenthcentury. This method of generating public revenues, however, was understood as far back asancient Greece, as illustrated by the following (Aristotle,The Politics, as quoted by Deans-Smith, 1992): “The way to make money is to get, if you can, a monopoly for yourself.Hence we find states also employing this method when they are short of money: they securethemselves a monopoly.”

In this section, I consider various historical and modern examples of mercantilistic finance,and show that these did indeed have a negative effect on technological development. Sucha survey necessarily must be brief, and I confine my observations to two general areas:Europe between the fifteenth and nineteenth centuries and less developed countries in ourown century.

Under the Hapsburgs and Bourbons, the Spanish empire made extensive use of monopoliesto raise revenue in the new world. While the crown monopolized many goods, includingmercury, salt, gunpowder, and snow, the tobacco monopoly was by far the most important.At its height in the late eighteenth century, monopoly profit from tobacco constituted nearly aquarter of all revenue generated in New Spain (Mexico). This revenue, moreover, bypassedthe regular treasury and went directly to the king (Haring, 1952, p. 305). The monopolyhad its own military force to discourage any form of competition, which may accountfor the near-total lack of technological innovation and consistent problems with qualityover the fifty years of its existence (Deans-Smith, 1992, especially chap. 5). Spain alsomonopolized, or attempted to monopolize, trade with its North American colonies. Untilthe end of the seventeenth century, the only legal trade was via massive flotillas approved bythe Spanish crown and organized by the wealthy commercial houses of Seville and Cadiz.This system generated crown revenues through direct duties and services in kind but exertedan important drag on innovation in the New World. As a noted historian (Haring, 1952,pp. 320–321) states:

Within a few decades (theconsulado] became virtually a closed corporation of afew great commercial houses enjoying a monopoly of traffic between Spain andits overseas empire.. . .Through it [the Seville merchants] also made loans andextended other financial aids to the crown, often under duress or to secure special

MERCANTILISM AND MODERN GROWTH 65

favors. . . . It was one of the most serious obstacles to [the colonies’] growth inindustry, in population and in general well-being.

The effect on Spain itself was hardly less depressing, and the monopolistic trading policywas part of a general economic and political decline that left the once-powerful country inruins by the end of the eighteenth century.

Mercantilism reached its zenity, however, with France and England in the seventeenthand eighteenth centuries. Their histories are painstakingly detailed by Heckscher (1935) inhis classic book. There he shows how the French monarchy, directed by the power Colbert,actuallystrengthenedthe medieval guild system, propping up monopoly rights for manyprofessions in exchange for revenue (p. 142, p. 178). In many cases, this revenue tookthe form of buying back the offices of newly created “inspectors” who had the power ofregulation and oversight over the particular craft guild (p. 178). Many times, these officeswere sold repeatedly, generating a stream of revenue (p. 180). The quote at the beginningof this article shows that, despite Heckscher’s interpretation of mercantilism as a guidingprinciple along five dimensions, he clearly saw that “fiscalism” (the raising of revenue) wasextremely important. Herbert Heaton (1937) went so far as to claim that Heckscher waswrong not to elevate public finance to as lofty position as his other five principles.11

It was also clear that the French monarchy actively reduced familiarity to help the monop-olistic guilds it supported. The starkest case is that of printed calico cloth. The techniquesfor making this cloth, which became extremely popular by the end of the seventeenthcentury, were more advanced in England and India, with the result that the producers oftraditional fabrics in France stood in danger of losing considerable profit. As a result, theFrench monarchy banned theuseof all such fabrics under severe penalty.12 This led to totalstagnation of this critical line of industrial development and the emigration of many Frenchcraftsmen (Heckscher, 1935, pp. 170–174; Ekelund and Tollison, 1981, pp. 93–96). Thereare many more examples, including gunpowder, salt, and luxury items, where the Frenchmonarchy actively supported monopoly and put a damper on technical innovation (Ekelundand Tollison, 1981, pp. 78–80).

Mercantile England, especially before the civil wars in the 1640s, demonstrated similartendencies with respect to monopoly creation and government revenue generation. Al-though the crown was never completely fiscally independent in France, it came closer toachieving that ideal than its English counterpart. Parliament had greater power than anyFrench institution to limit the King’s revenues, especially after 1688, forcing him to searchfor alternate sources of income.13 This conflict gave rise to the large trading companies likethe East India Company and the Hudson Bay Company, which owned their exclusive char-ters to the crown and ended up paying handsomely for them. Although the transfers weremade in many ways, the essence of the process was clear (Heckscher, 1935, pp. 439–441,emphasis added):

To say that the participation of the English kings in the undertakings was confined toattempts at shareholding is to draw rather artificial distinctions. There was no realdifference between this kind of participation and the other attempts to force moneyfrom the successful companies. Sometimes it was called participation, sometimesloans, occasionally gifts, and sometimes, though rarely, taxes.. . .The system thusinvolved an indirect taxation of consumers’ goods in the financial interests of the

66 MCDERMOTT

state. It was an indirect taxation of consumption by means of a monopoly, not inthe hands of the state but wielded by private individuals.

Whatever the theoretical underpinnings of mercantilism, in practice a very important com-ponent was the ability to tax the citizenry in a roundabout way via private, crown-charteredmonopolies. In 1621 there were about 700 different monopolies in England, including thosefor soap, iron, glass, dyes, printed books, and other key goods in fledgling industries (Hill,1961, pp. 32–34). There can be little doubt that these monopoly organizations exerted atremendous drag on technological innovation (North and Weingast, 1989), especially con-sidering that even the monopoly patents and charters were insecure, both from acts of ahostile Parliament and from revocation and resale by the king himself (Hill, 1961, p. 35).

Subsidies, especially export bonuses, also played an important role in mercantilist policyof the seventeenth and eighteenth centuries. At first glance, such payments would appear tocontradict the hypothesis that mercantilism was essentially a means of filling the treasury. Arecent paper by Irwin (1991), however, makes the point that, at least for the East India trade,Dutch managerial contracts acted like subsidy payments, and succeeded in diverting profitaway from the English. He also notes that direct subsidies could have been imposed byeither side to yield even larger profit—and state revenue—in the long run (pp. 1310–1311).To what extent other export bonuses of that era reflected such strategic trade policy is animportant question.

The key question formoderngrowth, however, is this: Does mercantilism still exist? Inmany developing nations, under one-party rule or managed by huge bureaucracies guidedby industrial policies, the answer is undoubtedly, yes. On the other hand, in moderneconomies, where tax reporting and collection are fairly efficient, mercantilism is much lessin evidence. Yet even in the United States, given the undue influence of certain lawmakers,it has been commonplace for industries and corporations to receive favors that enhance theircompetitive edge over others (some of which reduce familiarity) in return for payments,usually in the form of campaign contributions. A case in point is the ADM Corporation,which has enjoyed a sugar quota and ethanol subsidies that have earned it vast sums over theyears. In return, it funnels much smaller amounts into the coffers of various politicians. TheJones Act establishes a monopoly on shipping that is estimated to cost the U.S. economy$2.8 billion; maritime unions gave only $1.8 million to congressional candidates to ensurethe continuation of the monopoly (Wall Street Journal, 1998). Anther example concernstobacco in the United States. A good case can be made that the proposed deal was, before itscollapse, a simple exchange of legal immunity (a monopoly right) for a stream of revenue.The government was quite open about its need for the revenue that it expected to receive.The familiarity consequences of the tobacco deal may be minor, but one could argue thatthe decline of ship-building, and the distortions from ethanol in fuels do have negativeconsequences for technological development.

Governments of poor nations today are vastly more complicated than they were in theseventeenth and eighteenth centuries. Consequently, it is much harder to match any par-ticular mercantilistic policy with the revenue sent to the government. In many cases, suchpolicies are instituted after intense lobbying and accompanied by secret payoffs. In others,there is no overt policy, but corrupt functionaries are able to extract a stream of revenuefrom firms in exchange for selectively enforcing regulations or bid rigging. These have the

MERCANTILISM AND MODERN GROWTH 67

effect of establishing monopoly positions (Shleifer and Vishny, 1993). The result of suchpractices on familiarity may be large and negative.

Drawing an analogy between monarchies during the mercantile era and dictatorships inthe modern era, we might expect that present-day monopoly policies would be most visiblein countries ruled by strong men or single parties. Consider, for example, Trujillo in theDominician Republic, Somoza in Nicaragua, Marcos in the Philippines, and the Duvaliersin Haiti (Robinson, 1997). It is astounding how much of the economy each of the dictatorswas able to control, usually directly or through close relatives and henchmen. Trujilloowned 60 percent of the land and had claim to over half the gross domestic product of theDominican Republic; Somoza had perhaps half that in his country. Marcos had monopolycontrol of several key industries, and the Duvaliers likewise amassed great wealth, not justin the form of Swiss bank accounts but also in the form of ongoing interest in key industries.Needless to say, none of these countries was open to any form of competition that mightcurtail the profits of the monopolies owned and operated by the rulers. It is also truethat their comparative economic performance was dismal: the Philippines fell markedlybehind nations it had led just decades before, Nicaragua had the lowest growth rate in LatinAmerica, and per capita income in Haiti actually declined year after year.14

Democracies in the developing world tend to cloak mercantile tendencies in the man-tle of industrial policy. After World War II, governments of lesser developed countriesgrew gigantic under the influence of the idea that import substitution would result in fastindustrialization and catch-up to world leaders (Bruton, 1998). This process created enor-mous bureaucracies and severe price distortions and led to an interdependence of privatebusinesses and bureaucratic managers who relied on one another for gain in the form ofrevenues and political influence (Kruger, 1993). Monopolies came into existence and didin fact generate considerable revenue for privileged license-holders, which translated intorevenue and power for the bureaucrats that created and sustained them, and stifled competi-tion and learning (Krueger, 1993, especially chaps. 2, 4, 5). One example is the Informaticamarket reserve policy of Brazil, begun under the military regime in the late 1970s andphased out in the early 1990s. This law severely restricted both the importation and do-mestic production of personal computers by foreign firms. When the restrictions began,there were few domestic firms in a position to produce personal computers and peripherals.Charging prices from double to five-fold for two-year-old reverse-engineered technology(Schmitz and Cassiolato, 1992), it is likely that the profits of these firms were exceedinglyhigh. Although more firms entered the market, the initial advantage to the original produc-ers made them in practice a kind of monopoly. The opposition to this law from users ofcomputer technology became increasingly strong throughout the decade, as it became clearthat the law kept familiarity low and made it difficult for producers to take advantage of newtechnologies being developed abroad. What is less clear is the nature of the transfer back tothe government as payment for granting the privileged position. Since such payments aregenerally illegal in modern democracies and thus hidden, it is difficult to form tests of themercantile hypothesis. Still, until we have evidence to the contrary, it is difficult to believethat the large concessions granted over oil, minerals, and the large state manufacturing andfinancial enterprises in Brazil, Mexico, Argentina, Turkey, and India, were the result ofdisinterested bureaucrats uninfluenced by politicians who stood to earn real transfers.15

68 MCDERMOTT

6. Government Policy andκ

Governments face a tradeoff that arises from the existence of two sources of revenue. On theone hand, a lowκ is good for monopoly and keepscurrentrevenue high. On the other hand,a highκ results in lower immediate revenues but does generate abnormally high short-termgrowth and a permanently higher path for real income, which raises revenuesin the future.

The basic nature of the tradeoff facing the government can be understood with referenceto Figure 2. Assume that two countries initially possess the sameκ and the sameh0, so bothbegin at point B. Now, let one of them succeed in raising itsκ, while the other maintainsthe original value. The rise inκ would reduce work effort by (8) causing areductionin percapita incomey0. Since people devote more effort to learning, the economy’s output falls,reducing government revenue. The low-κ country will eventually converge to the low-ypath, while the other country, which suffers an initial decline iny, will converge to thehigh-y path, generating a high level of tax revenue. Both grow at the same rate in the longrun, but the consequences for the government budget are likely to be very different.

If familiarity were easy to change, the best policy for the government would be to followthe time path ofκ that maximized the integral of discounted public revenue from the presentto the infinite future. This policy would be time-consistent and would induce just the righteffort from individuals to produce the proper balance between current and future revenue.There are two problems involved with this approach. First, finding such a path forκ is quitedifficult, even with the relatively simple structure of the model of this article. Second, andof greater importance, changing national familiarity is likely to be so difficult politicallythat it can be done only infrequently. To take one example, Japan’s decision to becomemore open after the Meiji restoration followed centuries of a consistent policy of closureand lowκ. In the modern era, Chile became much more open and liberal in the 1970s, butonly following a traumatic political upheaval. Argentina reversed decades-long Peronistpolicies in the 1990s to become more outwardly oriented but did so only after running theeconomy into the ground and losing a war to England.

To reflect the rigidity in picking familiarity, I assume that the government decides on aconstantvalue ofκ over a finiteplanning horizonof lengthT . We may think of the planninghorizon as that period over which the government expects to earn the revenue generatedby the level of familiarity that it sets. For some societies, like the United States and mostof Western Europe,T may be quite long, since the government is greater than the politicalparty that happens to be in power. For other nations, such as Haiti, the planning horizonmay be very short, since there is a much stricter identification of government revenue withparty income.

Although the government usesT to setκ, I assume that opportunities to changeκ mayarise before the end of the planning horizon. If such an opportunity occurs, the governmentpicks a new level of familiarity and again bases the decision on the full planning horizonT .By distinguishing between the planning horizon and the more frequent, but random, abilityto adjust familiarity, I combine forward-looking behavior with some degree of flexibility.

Let the government selectκ to maximize the revenue integral:

J =∫ T

0R(t)e−ρGtdt, (19)

MERCANTILISM AND MODERN GROWTH 69

which can be expressed as16

J(κ,V, F0) =∫ T

0[τ +mσ(1− κ)]Ä(t)2 F(t, F0, κ)

1+ 1γ

κγ X(t, F0, κ)N(t)e−ρGtdt. (20)

In these expressions,T is exogenously given,V refers to the vector of fiscal parameters—[τ,m, σ ]—particular to the country in question,ρG is the government’s discount rate, andF0 is the initial state faced by households.

The functionsF(t, F0, κ) andX(t, F0, κ) refer to the values ofF andx along the optimalpath QA in Figure 1. I assume that the household discount rateρ and the learning spillovereffectγ are the same across countries, so the paths ofF andx differ only to the extent thatκ and the initial valueF0 differ.

World technologyÄ and the country’s populationN grow exogenously. Normalizingtheir initial values to 1 allows us to simplify the revenue integral to

J = [τ +mσ(1− κ)]∫ T

0

B(t, F0, κ)

κγe−r t dt, (21)

whereB(t, F0, κ) is the ratio toF to x, and the government’s effective rate of discountr isadjusted for the growth in populationη and technology:r ≡ ρG − η − 2gÄ.

It is not possible to find an analytical solution to the problem of maximizingJ, since thereis no closed-form expression for theB(.) function. Instead, I use numerical methods thatbuild on the information in Table 1. Using the parameter values given there, it is possible tofind a numerical representation of the QA path orpolicy function. Using the policy functionand the differential equation forF we can then find numerical approximations for the twofunctions that formB(t, F0, κ).17

6.1. Initial Calibration and Time Consistency

For different values ofκ—given the other fiscal variables and the starting endowmentF0—individuals would choose distinct time paths forF andx, generating different values forgovernment revenue in (21). After evaluating the integral numerically for each value ofκ,the government then picks the familiarity level that maximizes its total revenue. Even ifmor σ were zero, so the government had no interest in setting up monopolies for financialreasons, it might still be in the government’s interest to close the economy if the planninghorizonT were sufficiently short. The reason can be seen by reference to Figure 2: byreducingκ the government raises current work effort, and thus tax revenue, at the expenseof learning effort. IfT is small enough, even ifmσ = 0, it may be rational to setκ < 1since future tax revenues do not matter. Setting the parametersτ = 0.10, T = 75, andρG = ρ = 0.04 (which means thatr = 0.005, once we subtract population growthη = 0.015 and technology growth 2gÄ = 0.02) avoids this outcome. For those parametervalues, if eitherm = 0 orσ = 0, the government would always setκ = 1, no matter whatthe level ofF . In other words, in the absence of mercantilistic influence, government policywould be both time-consistent and optimal for the households.

70 MCDERMOTT

Let the function

κ = κG(F;mσ, τ), (22)

be the function that relates the fiscal parameters and the state of the economyF to thegovernment’s choice ofκ.18 The vertical line at 1 in Figure 3 labeledκG,0 shows the valueof κ that the government would set, for different values ofF , given thatmσ = 0, andτ = 0.10. The fact that it is vertical means that for any level ofF , people behave in such amanner that the government finds it rational to keep the economy completely open:κ = 1.The upward-sloping line labeledF∗(κ) in Figure 3 refers to the steady-state level ofFas given by (13). The two lines cross at point E, which is the steady state when there isno mercantilism. Assume that the economy began withF = 1, so that individual humancapital were equal to the level of world technology. In Figure 3 the government places theeconomy at point D, since we know that withmσ = 0 the best choice for the governmentis κ = 1. With complete openness,F grows over time. Although the government getsopportunities to reoptimize and changeκ, it will not do so. The economy will convergedirectly to point E, and its relative income level eventually will be the equal of the worldleader.

Before we deal with mercantilism, it is important to see what would happen if the planninghorizon were very short. Numerical methods reveal that whenmσ = 0 the policy problemis of the bang-bang variety: ifT is long,κ = 1 is optimal; ifT is short, then it is optimal tosetκ at itsminimumlevel, which I have set at 0.06. This value was chosen because relativeincome of 6 percent of the U.S. level reflects the state of some of the world’s poorest nations,like Haiti and Bangladesh.19 It is, therefore, a possibility that some very closed economiesare that way not because of mercantilistic influence but because they are controlled byregimes that have little stake in the future of the country. This may be because they belongto certain ethnic groups that will be driven from all power if they lose an election or fail toput down a rebellion.

6.2. Mercantilism and Familiarity: The Baseline Case

Monopolies may be strong enough to make mercantilism worthwhile for the government. Asa baseline, assume that the maximum transfer rate isσ = 0.50, so that if the economy werecompletely closed(κ = 0) monopolies could skim half of the household’s income beforethe government took its cut. Further, assume that the mercantilistic premiumm= 0.30 sothe government can earn four times more revenue per dollar by taxing monopolies instead ofhouseholds. As above, the government planning horizon isT = 75 and the traditional taxrateτ is kept at 10 percent. Let the country again begin with both human capital and worldtechnology set to 1, so the initial value ofF is also 1:F0 = 1. The average citizen of thiscountry would have a low income, but his effort-productivity for accumulating more humancapital,κγ /F , would be relatively high. The upper half of Table 2 shows the complete setof fiscal parameters.

For this case, the government revenue integral reaches a maximum for a familiarity valueof κ = 0.80. Complete openness is not in the government’s interest. The actual transfer rate

MERCANTILISM AND MODERN GROWTH 71

Figure 3. Familiarity dynamics.

to monopolies from household income iss= σ(1− κ) = 0.10, and the fraction of incomeretained by the household is(1− τ)(1− s) = 0.81. Finally, total discounted consumptionover the seventy-five-year planning horizon withκ set to 0.80 is approximately 70 percentof the total consumption that would have been realized with full familiarity(κ = 1). Thisis called the “consumption shortfall ratio” in the bottom half of Table 2. In Figure 3, pointH corresponds to the initial state in the baseline case. The locus labeledκG,1 representsequation (22), the rational familiarity level, for the baseline parameters.

6.3. Familiarity Dynamics

SinceF is below its steady stateF∗(κ) at point H, it will be growing over time. As it grows,the government is presented with randomly timed opportunities to change openness. Totake an example, it takes about twenty-three years forF to reach 5 (point I) at which time, let

72 MCDERMOTT

Table 2.Baseline fiscal parameters and results.

Variable Symbol Value

ParametersTraditional tax rate τ 0.1Mercantilistic tax premium m 0.3Maximum monopoly transfer rate σ 0.5Government planning horizon T 75Government net discount rate r 0.005Initial human capital h0 1.0Initial state F0 1.0

Results: Initial stateFamiliarity κ 0.80Monopoly transfer rate s 0.10Income retention share (1− τ)(1− s) 0.81Consumption shortfall ratio 0.70

Results: Steady stateFamiliarity κ 0.20Monopoly transfer rate s 0.40Income retention share (1− τ)(1− s) 0.54Consumption shortfall ratio 0.34

us say, the government is presented with the opportunity to pass a new policy with respectto openness. It will actually choose toreduceκ to 0.65. This is shown as point J, whichlies along theκG,1 locus. It may appear counterintuitive that an economy that accumulatesmore human capital relative to world technology, and thus achieves a greater standard ofliving, actually finds it in its interest to close itself off to a greater degree from the rest ofthe world. We must keep in mind, however, that such a policy is instituted not because it isin the citizens’ interests but because it is in the interest of the government. It is rational toclose the country because the higherF means that extra work—accomplished by reducingκ—yields greater income passing through monopolies and available for taxation.

Human capital continues to rise relative to world technology in spite of the reducedopenness, and if the next opportunity to adjustκ occurs whenF reaches 10, the governmentwould reduceκ to 0.41. This is shown as point K. The steady state occurs at point L, whereF = 13.88 andκ = 0.20. There is, however, no guarantee that the economy will convergeto the steady state, since to do so requires that an opportunity to adjustκ arise just exactlywhenF reaches 13.88. If, as appears likely,F exceeds that level before the government canchange familiarity, the resulting downward adjustment inκ would be so large thatF wouldbegin to decline. In other words, there may exist acycle in κ around the steady-state L,one of uneven amplitude based on the infrequent, random events that give the governmenta chance to adjust.20

If the steady state were achieved, monopolies would seize a full 40 percent of the incomeof households that, after the government takes its cut, would retain only 54 percent of whatthey earned (see Table 2, last section). With familiarity of 0.20, the economy’s per capitaproduct would be afifth of that of the world leader. Nations in 1990 with relative percapita income in the range of 15 to 25 percent of the U.S. level include Argentina, Brazil,Columbia, Poland, Thailand, Turkey, and Yugoslavia.21

MERCANTILISM AND MODERN GROWTH 73

Table 3. Steady-stateκ:effects ofτ andm.

m

τ .05 .10 .30

.10 .59 .35 .20

.20 1.0 .59 .27

.30 1.0 .82 .35

Cycles generated by governments lacking fine control may demonstrate deep downswingsin κ, since there is a discontinuity in theκG locus due to the inherent structure of the model.WhenF gets above point M in Figure 3, for example, the best choice for the government isthe minimum level of familiarity.22 If the country were this closed, accumulation of humancapital would be far inferior to the growth in world technology, andF would sink directly.At the next opportunity, the government would raiseκ back to the appropriate place alongκG,1.

The notion that cycles may characterize equilibrium is attractive. The postwar experienceof Latin America provides a good example of a region that has lacked consistency withregard to openness. It was closed for several decades and then, in response to stagnatingrelative incomes, took decisive measures to become more liberal with respect to capital andtrade. Chile is at the forefront of the current wave of raisingκ, but one could also point toArgentina, Brazil, and Colombia. Others, like Venezuela and Peru, remain quite closed.

6.4. The Importance of Raising the Traditional Tax Rate

Mercantilistic forces lead to closed economies, and the strength of that closure appearsto rise with development. But are there any countervailing forces that drive an economytowards openness? The answer is yes: as the traditional tax rateτ rises, the governmentvoluntary raisesκ and reduces its reliance on monopolies as a source of public finance. Wehave good reason, furthermore, for expectingτ to rise withF . The greater isF , the larger isthe human capital of the average citizen, and the larger is the middle class. Since a greaterfraction of income can be obtained from the middle class than from either the poor or therich, it is reasonable to assume thatτ rises with the state of development.

Whenτ rises, theκG locus shifts to the right. Ifτ doubled toτ = 0.20, for example, thegovernment would select a familiarity level of 0.27 in the steady state. This is noted in thelast column of Table 3. If the tax rate rose to 0.30, which might reasonably considered amaximum, the locus would move to the position markedκG,2, and the steady-state familiaritylevel would rise to 0.35 (point N in Figure 3). Although this represents a real improvementin living standards, it still leaves the economy far below the world leader in the permanentdistribution of world income.

74 MCDERMOTT

6.5. Full Convergence

Full convergence cannot take place with higher taxes alone. At some point, the degree ofmercantilism must fall or the economy will continue to lag in relative income. A declinein m or σ also moves theκG locus to the right, leading the government to institute policiesthat open the nation to foreign competition and ideas.23 As noted in the second columnof Table 3, a decline inm to 0.10 raises familiarity significantly: if the tax base is high(τ = 0.30), the government raises familiarity all the way toκ = 0.82. The steady stateis shown as point P alongκG,3 in Figure 3. This number is consistent with the income ofnations like Australia, Germany, Japan, and Sweden (relative to the United States), all ofwhom have both higher tax rates and a greater degree of state collaboration with industrythan does the United States.

Does full convergence requirem = 0? As shown in the first column of Table 3, theanswer is no. The government would voluntarily open the economy completely and selectκ = 1 if m = 0.05 andτ = 0.20. The latter figure is about the same as the average U.S.tax rate today. It is more difficult to measure the mercantilistic premium. In fact, there isno reason to believe, just because the United States leads the world in per capita income atthis point, that it is completely open. It is more reasonable to assume that U.S. familiarityfalls short of 1 but remains higher than that of other nations.

7. Conclusions

Even though a policy of openness maximizes individual welfare, raises short-run growth,and allows laggards to catch up to world leaders, such a policy is often not chosen by thosein political power. Until we understand why not, we will not be able to fully account forthe fact that convergence in living standards around the world is occurring at a very slowpace, if at all.

The explanation advanced here involves the concept of mercantilism, defined to be thecollaboration between the government and private monopolies for the expansion of publicrevenue. By establishing monopoly rights for a certain firm and then siphoning off a partof its profit, a government may be able to increase its revenue beyond what is attainablewith traditional sources. To make the monopoly right effective, however, it must closethe economy to new ideas. The key familiarity parameter is reduced by conscious publicpolicy. A serious consequence of this closure is a reduction in the ability of the citizenry toenhance human capital and provide economic growth.

I have shown that there is considerable support in the literature for the idea that mercan-tilism, in spite of the rhetoric, was primarily a system of fiscal enhancement. The formalmodel showed that two countervailing forces arise to change familiarity over time. On theone hand, the growth in human capital generates a tendency for familiarity to be reduced.On the other hand, a rise in the traditional tax rate tends to cause the government to raisefamiliarity. Nonetheless, if nothing can be done to reduce the degree of mercantilism, fullconvergence will remain elusive.

MERCANTILISM AND MODERN GROWTH 75

Appendix

Necessary Conditions for the Households

The Hamiltonian of this problem—after normalizing the initial population level at 1—isthe following:

H = ln c+ qh[h1−γ (κÄ)γeL

]+ λ1(ω h eW − c)+ λ2(1− eW − eL). (23)

The middle constraint on consumption comes from (16), whereω < Ä because of taxes.The following first-order conditions arise:

1

c= λ1, (24)

λ1ω h = λ2, (25)

qhhκγ

F= λ2, (26)

and, as in the textF = ( hÄ)γ . Eliminate theλi from these conditions, and use the constraint

on consumption (16) to obtain (8).The costate must move according to

qh = ρqh − ∂H

∂h. (27)

To obtain (10), begin with the definitionx ≡ qhh, and use (27) and (3). Differentiate (23)and use (24) to eliminateλ1.

Numerical Representation of the Time Paths

Equations (10) and (11) form a nonlinear dynamic system in(F, x), given values of pa-rameters, includingκ. The first step is to find a numerical estimate for the policy functionrepresented by the QA path in Figure 1 withκ = 1. To do so, I employed the time-elimination method (Mulligan and Sala-i-Martin, 1991). Call this numerical representationx = v(F). Thegeneral policy function, which is relevant for different values ofκ, is givenby

x = P(F, κ) = v(F/κγ ). (28)

This simplification is possible because the entire system could be cast in terms ofE ≡ F/κγ ,in which case the motion and steady-state levels would be invariant with respect toκ.24

The general policy function, to emphasize, is a numerical approximation and does nothave a simple closed form.

The next step is to find time paths forF andx. To do so, I first substituted the policy func-tion for x into (11), setκ = 1, and solved the differential equation numerically beginningat F0 = 0.01. Call the result thebaseline time path F= κ(t). The other baseline time path

76 MCDERMOTT

is x = L(t) = P(κ(t),1.0). These are the values of the state variables at any time, giventhatF begins at the very low value of 0.01, and familiarityκ has the maximum value of 1.0.The baseline path is also the time path forE, defined above. This is very important becauseit allows us to generalize the paths to apply to any initialF and any givenκ. We can easilygeneralize from theE = F = κ(t) path because a change inκ changes the initial value ofE but otherwise does not disturb the path. Therefore, consider the timetz it takes forE toreach the valueF0/κ

γ . This time can be found by numerical methods. Call ittz = Q( F0κγ).

It then follows that in generalE = κ(t + Q( F0κγ)). Finally, to find the time path forF for

any values ofF0 andκ we multiply the last result byκγ :

F = F(t, F0, κ) = κ(

t + Q

(F0

κγ

))κγ . (29)

To find x at any time, substitute (29) into the general policy function (28) to get

x = x(t, F0, κ) = P(F(t, F0, κ), κ). (30)

These two functions correspond to theF(·) andX(·) functions in the government’s revenueintegral (20) and appear as the ratio in theB(·) function in (21).

The two time paths are strictly numerical. That is, they are sequences of short functionsspliced together to span the range that was specified. Nevertheless, once values for the fiscalparameters(τ,m, σ ), the initial condition(F0), and familiarityκ are specified, equation(21) can be integrated numerically to find a value for government revenue. It is also possible,using numerical methods, to form a graph ofJ as a function ofκ, and to maximizeJ withrespect toκ.

To construct Figure 3, I began withF0 = 1, maximizedJ to find κ; then used thatκ tosee how long it would takeF to reach 5. At that time (about twenty-three years) I thenusedF0 = 5 to begin the process anew. Iterations proceeded until the steady state at L wasfound. This procedure was used repeatedly to fill in the values in Table 3.

All numerical routines were run withMathematicaversion 3.0.

Acknowledgments

I would like to give special thanks to Marvin Goodfriend for his important contributionsto this work. Two referees were instrumental in improving this article. The participantsof the University of South Carolina Economics Department seminar series, especially LisaRutstrom, Bill Phillips, Chip Chappell, and Kin Blackburn, also provided very helpfulcomments.

MERCANTILISM AND MODERN GROWTH 77

Notes

1. The first proposition has been supported in a series of papers by Quah (e.g., 1996, 1997) and in the work ofParente and Prescott (1993), Jones (1997), and Pritchett (1997). The second proposition has been investigatedfor a long time. Although much of this work finds a positive relation between trade and growth (e.g., Feder,1982; Edwards, 1993; and Harrison, 1995) the work is subject to criticism for not dealing adequately withsimultaneity. That problem is addressed in the paper by Frankel, Romer, and Cyrus (1996), who use a gravityequation to produce instruments for the measure of openness. Ben-David (1996) shows that groups of nationsthat trade with one another also converge in terms of per capita GDP, which is consistent with the idea thatopenness enables lagging countries to catch up to leaders within the group. In work that is directly related to thepresent article, Hall and Jones (1998) show that countries that speak an international language are more likelythan others to have high GDP per worker. Language, unlike the trade share, is an exogenous (if imperfect)measure of openness to international ideas.

2. The termmercantile systemmay have originated with Adam Smith (1776, Book 4), who compared it unfa-vorably to classical laissez-faire. It was a hundred years later that Gustav von Schmoller (1886) used the termmercantilismin his analysis of national unification in the context of Frederick the Great’s policy of wrestingcontrol of the economy from cities and towns in Prussia. The classic work on mercantilism is Eli Heckscher’s(1935) lengthy treatise.

3. Jacob Viner (1937, p. 115) noted that “the bulk of the mercantilist literature” was nothing more than special-interest pleading, often disguised as disinterested policy analysis. Heckscher’s (1935) and Heaton’s (1937)quotes at the beginning of this article shows that both were well aware of this aspect of mercantilism.

4. Other theoretical works that draw the link between knowledge and growth in the international economy includeGrossman and Helpman (1991), Barro and Sala-i-Martin (1997), and Eaton and Kortum (1996).

5. It is impossible to ignore the fact—emphasized by Romer (1990, 1996)—that many technical ideasaresufficiently proprietary that they cannot be freely used everywhere (so thatÄ is not identical across countries).In this article, I do not deal with the ability to exclude others from technical ideas.

6. Equations (1) and (2) are adapted in simplified form the two-country model of Goodfriend and McDermott(1998). A more precise account is given there of the determination of technologyÄ and the effect of familiarityκ on the formation of human capital.

7. One way to justify this assumption is to assume that spending on productive public goods is matched byspending on destructive goods, both for rent seeking and war, in such a way that there is no net effect oncurrent output. Rosenthal (1998) places war spending at the heart of his model. McGuire and Olson (1996)construct a model in which the desire for revenue is tempered by the need to provide public goods and thedead-weight cost of taxation. Their main focus is on the level of taxation in a static economy under differentpolitical structures. Robinson (1997) recognizes that the state may decide not to provide public goods becauseit makes political change less costly. As here, in his model the state chooses an inefficient outcome to maintainits own wealth at the expense of the population as a whole.

8. This assumption is a simplification. As suggested by a referee, one could think of this rate as one that maximizesrevenue in Laffer-Curve fashion. It is reasonable to assume that this rate rises with the level of development.Even so, there are exceptions to the proposition that high direct taxation requires high development. Peter theGreat was able to extract an extraordinarily high fraction of income directly from bonded peasants. Althoughhis economy was at a very low level of development, he had more power than most rulers of the mercantileor modern period (Mavor, 1925, chap. 6). In practice,averagetax rates shows little variation in developednations: the average tax rate in the United States has been steady at about 20 percent for decades (EconomicReport of the President1997).

9. Parente and Prescott (1998) develop a much richer structure to explain why monopolization is detrimental tooutput. They do not consider openness or growth explicitly, but their underlying mechanism—a failure toadopt superior technology—is similar to that here.

10. Although the mercantilistic tax premium is exogenous here, a referee pointed out that one could look at thisparameter as the result of a Nash bargaining process between monopolies and the government. An elaborationof this process would allow us to relate the value ofm to other exogenous variables.

11. For the record, Heckscher identified the following guiding principles of mercantilism: unification, nationalpower, protectionism, money and trade, and social concept.

78 MCDERMOTT

12. Heckscher’s claim that 16,000 people were put to death over calico smuggling and production has been disputedby Ekelund and Tollison (1981, p. 94).

13. The conflict between the monarch and an elite (represented by Parliament in England and the Estates Generalin France) is modeled as a game of fiscal control to seek fortunes via war by Rosenthal (1998). His approachemphasizes, as I do here, that it is impossible to separate efficiency and institutions from the fiscal desires ofrulers. According to Ekelund and Tollison (1981, p. 69), Parliament was not as mercantilistic in its policychoices because the costs were greater for a democratic institution than for a monarch.

14. The foregoing follows Robinson (1997) closely. He cites, among others, Vedovato (1986) and Wiarda (1968)for Trujillo, Rempel (1993) for Marcos, Crawley (1979) for Somoza, and Lundahl (1992) for Haiti.

15. Southeast Asia presents an intriguing case, with no consensus about what caused its generally high growthfrom the mid-1970s to the mid-1990s. South Korea and Taiwan appear to corroborate the importance offamiliarity, since they opened considerably just before their growth took off (Bruton, 1998). Still, they, likeJapan, relied on industrial policy more than is often acknowledged. All three appear to have found a way toraiseκ even as their governments maintained control over investment. This may have been true also, thoughto a smaller extent, of Indonesia and Malaysia, who, in spite of crony capitalism, managed high growth duringpart of the period. Recent developments in the latter, however, make us realize that it is too early to tell if theirsuccess will be long lasting.

16. To derive (20), substitute (18) forR, and subsequently (1) fory. Following this, use (4) to eliminateh and (8)to eliminateeW.

17. See the appendix for an explanation of how the policy function and time paths are found.

18. By “state of economyF ,” I mean the value ofF that exists whenever the government makes its decision overκ. As noted earlier, this will happen at discrete, random times.

19. Although some strife-ridden nations in Africa go as low as 2 percent of U.S. GDP, I do not set the minimumκ

that low for technical reasons. Whenκ is that low, it is possible that no steady state exists for some otherwisereasonable parameter values.

20. Mancur Olson (1982) emphasized the fact that nations face discrete opportunities, which may look likedisasters, to liberalize both internally and externally.

21. All of the numbers in this section come from Summers and Heston (1991).

22. ForF above M, there may exist alocalmaximum to the government’s objective function away from the corner,but if so, the value of the revenue integral there falls below the value provided by the minimumκ.

23. The government does not care if its mercantile income rises because the monopolists have succeeded in raisingthe transfer rate from households (a greaterσ ) or because the government itself is better able to obtain morerevenue from the monopolists (a largerm). As we see from (21) only the productmσ is important. In Table 3,then, we could replace them-values withσ -values of the same proportional magnitude. From the point ofview of the private sector, however, it does matter, even before any changes in familiarity occur: monopolistsprefer a largerσ , while households prefer a largerm.

24. Both the lower and upper stable arms were found. In this appendix, however, I focus only the lower path.

References

Barro, Robert, and Sala-i-Martin, Xavier. (1997). “Technological Diffusion and Convergence.”Journal ofEconomic Growth2, 1–26.

Ben-David, Dan. (1996). “Trade and Convergence Among Countries.”Journal of International Economics40(May) 279–298.

Bruton, Henry J. (1998). “A Reconsideration of Import Substitution.”Journal of Economic Literature36, 903–936.

Crawley, Eduardo. (1979).Dictators Never Die: A Portrait of Nicaragua and the Somoza Dynasty. New York:St. Martin’s Press.

Deans-Smith, Susan. (1992).Bureaucrats, Planters, and Workers: The Making of the Tobacco Monopoly inBourbon Mexico. Austin: University of Texas Press.

Eaton, Jonathan, and Samuel Kortum. (1996). “Trade in Ideas: Patenting and Productivity in the OECD.”Journalof International Economics40, 249–278.

MERCANTILISM AND MODERN GROWTH 79

Economic Report of the President. (1997). Washington, DC: U.S. Government Printing Office.Edwards, Sebastian. (1993). “Openness, Trade Liberalization, and Growth in Developing Countries.”Journal of

Economic Literature31, 1358–1393.Ekelund, Robert, and Tollison, Robert. (1981).Mercantilism as a Rent-Seeking Society. College Station: Texas

A&M University Press.Feder, Gershon. (1982). “On Exports and Economic Growth.”Journal of Development Economics12, 59–73.Frankel, Jeffrey, David Romer, and Teresa Cyrus. (1996). “Trade and Growth in East Asian Countries: Cause

and Effect?” NBER Working Paper 5732.Goodfriend, Marvin, and John McDermott. (1995). “Early Development.”American Economic Review85,

116–133.Goodfriend, Marvin, and John McDermott. (1998). “Industrial Development and the Convergence Question.”

American Economic Review88, 1277–1289.Grossman, Gene, and Elhanan Helpman. (1991).Innovation and Growth in the Global Economy. Cambridge,

MA: MIT Press.Hall, Robert, and Charles Jones. (1998). “Why Do Some Countries Produce So Much More Output Per Worker

Than Others?” NBER Working Paper 6564.Haring, C. H. (1952).The Spanish Empire in America. New York: Oxford University Press.Harrison, Ann. (1995). “Openness and Growth: A Time-Series, Cross-Country Analysis for Developing Coun-

tries.” NBER Working Paper 5221.Hill, Christopher. (1961).The Century of Revolution. Edinburgh: Nelson.Heaton, Herbert. (1937). “Heckscher on Mercantilism.”Journal of Political Economy45, 370–393.Heckscher, Eli. (1935).Mercantilism. London: Allen & Unwin.Irwin, Douglas. (1991). “Mercantilism as Strategic Trade Policy: The Anglo-Dutch Rivalry for the East India

Trade.” Journal of Political Economy99, 1296–1314.Jones, Charles. (1995). “R&D-Based Models of Economic Growth.”Journal of Political Economy103, 759–784.Jones, Charles. (1997). “Convergence Revisited.”Journal of Economic Growth2, 131–153.Kremer, Michael. (1993). “Population Growth and Technical Change: One Million B.C. to 1990.”Quarterly

Journal of Economics108, 681–716.Krueger, Anne O. (1993).Political Economy of Policy Reform in Developing Countries. Cambridge, MA: MIT

Press.Lucas, Robert E., Jr. (1993). “Making a Miracle.”Econometrica61, 251–272.Lundahl, Matts. (1993). “Introduction.” InPolitics or Markets? Essays on Haitian Underdevelopment. New

York: Routledge.Mavor, James. (1925).An Economic History of Russia(2nd ed.). Vol. 1. New York: Dutton.McGuire, Martin, and Mancur Olson. (1996). “The Economics of Autocracy and Majority Rule.”Journal of

Economic Literature34, 72–96.Mulligan, Casey, and Sala-i-Margin, Xavier. (1991). “A Note on the Time-Elimination Method for Solving

Recursive Dynamic Economic Models.” NBER Technical Working Paper 116.Nelson, Richard, and Edmund Phelps. (1966). “Investment in Humans, Technological Diffusion, and Economic

Growth.” American Economic Review56, 69–75.North, Douglass. (1981).Structure and Change in Economic History. New York: Norton.North, Douglass, and Barry Weingast. (1989). “Constitutions and Commitment: The Evolution of Institutions

Governing Public Choice in Seventeenth-Century England.”Journal of Economic History49, 803–832.Olson, Mancur. (1982).The Rise and Decline of Nations: Economic Growth, Stagflation, and Social Rigidities.

New Haven: Yale University Press.Parente, Stephen, and Edward Prescott. (1993). “Changes in the Wealth of Nations.”Quarterly Review(Federal

Reserve Bank of Minneapolis), (Spring), 3–16.Parente, Stephen, and Edward Prescott. (1994). “Barriers to Technology Adoption and Development.”Journal

of Political Economy102, 298–321.Parente, Stephen, and Edward Prescott. (1998). “Monopoly Rights: Barriers to Riches.” Minneapolis Federal

Reserve Bank.Pritchett, Lant. (1997). “Divergence, Big Time.”Journal of Economic Perspectives11, 3–17.Quah, Danny. (1996). “Convergence Empirics Across Economies with (Some) Capital Mobility.”Journal of

Economic Growth1, 95–124.Quah, Danny. (1997). “Empirics for Growth and Distribution: Stratification, Polarization, and Convergence

Clubs.” Journal of Economic Growth2, 27–59.

80 MCDERMOTT

Rempel, W. C. (1993).Delusions of a Dictator. Boston: Little Brown.Rivera-Batiz, Luis, and Paul Romer. (1991). “Economic Integration and Economic Growth.”Quarterly Journal

of Economics106, 531–556.Robinson, James. (1997). “When Is a State Predatory?” University of Southern California.Romer, Paul. (1990). “Endogenous Technological Change.”Journal of Political Economy98, S71–S102.Romer, Paul. (1996). “Why Indeed in America? Theory, History, and the Origins of Modern Economic Growth.”

American Economic Review86, 202–206.Rosenthal, Jean-Laurent. (1998). “The Political Economy of Absolutism Reconsidered.” In Robert Bates, Avner

Greif, Margaret Levy, J.-L. Rosenthal, and Barry Weingast (eds.),Analytic Narratives. Princeton: PrincetonUniversity Press.

Schmitz, Hubert, and Jos´e Cassiolato. (1992).Hi-Tech Industrial Development: Lessons from the BrazilianExperience in Electronics and Automation. London: Routledge.

Schmoller, Gustav von. (1895).The Mercantile System and Its Historical Significance. New York: Macmillan.Schleifer, Andrei, and Robert Vishny. (1993). “Corruption.”Quarterly Journal of Economics108, 599–617.Smith, Adam. (1776; 1937).The Wealth of Nations. New York: Modern Library.Summers, Robert, and Alan Heston. (1991). “The Penn World Table (Mark 5): An Expanded Set of International

Comparisons, 1950–1988.”Quarterly Journal of Economics106, 327–368. Data at NBER.org.Vedovato, Claudio. (1986).Politics, Foreign Trade, and Economic Development: A Study of the Dominican

Republic. London: MacMillan.Viner, Jacob. (1937).Studies in the Theory of International Trade. New York: Harper.Wiarda, H. (1968).Dictatorship and Development: The Methods of Control in Trujillo’s Dominican Republic.

Gainesville: University of Florida Press.Wall Street Journal. (1998). Editorial. October 5.Young, Alwyn. (1993). “Invention and Bounded Learning by Doing.”Journal of Political Economy101, 443–472.